Abstract

Objective: We compared three measures of Intra-Individual Variability (IIV): Overall Test Battery Mean (OTBM) SD, CV, and Range on ability to classify individuals grouped by MMSE scores. Method: Archival data from 12683 individuals from the National Alzheimer's Coordinating Center's (NACC) Uniform Data Set (UDS) were analyzed. Demographics: 42.6% male, 82.6% Caucasian, mean age of 73.0 (SD = 9.3), 47.6% normal cognition, 23.6% Alzheimer's dementia (AD) or other non-AD disorders (23.6%). Individuals were grouped based on MMSE scores: 28–30 (n = 7737), 24–27 (n = 3223), 20–23 (n = 1226), 16–19 (n = 367), and 9–15 (n = 130). An OTBM for each participant was calculated by converting individual test performances to a T score and then averaging all T scores for each individual. We then calculated a standard deviation for each individual around his or her own OTBM (OTBMSD). CV was calculated by dividing OTBMSD by the OTBM for each individual. Range was calculated by subtracting each individual's lowest T score from his or her highest T score. Results: Spearman rho correlations showed pathology, as measured by MMSE scores, was correlated with OTBM rs(12683) = −.67, p < .001, OTBMSD rs(12683) = .43, p < .001, CV rs(12683) = .60, and range rs(12683) = .424, p < .001; greater neuropathology was associated with lower test performance and greater IIV. Kruskal-Wallis nonparametric tests revealed significant main effects for OTBMSD [F(4, 12683) = 2512.1, p < .001], CV [F(4, 12683) = 4569.2, p < .001], and range [F(4, 12683) = 2332.3, p < .001] on MMSE groupings. Conclusion(s): In this sample, CV was best able to differentiate between MMSE groupings across all groups. Range and OTBMSD were able to discriminate between groups with higher MMSE scores but not as well at lower levels of functioning.